Presentations


Introduction to Data Neural Networks

I am writing this paper to show the advantages of combining two important networks: a data network and a neural network. Data networks have been with us for decades, but now have evolved into the Internet with global connectivity. Packet data networks now carry virtually all forms of communications and can operate in excess of ten gigabits per second.

Neural networks have been with us since the dawn of time, and each of us has one attached to our bodies. Neural networks, however, have several limitations: In our bodies, at the cellular level, there is a limit to the number of neurons that can communicate with any given neuron. The limitation occurs because each neuron must directly connect to each neuron with which it needs to communicate. Also neural networks are very slow as they operate at the cellular level. People have been experimenting with artificial neural networks, but for the most part, they are analog and use special purpose components making them slow, expensive, and low density.

Data networks are very fast, but have no intelligence, are easily hacked, and require armies of knowledgeable workers to keep them set up and operating efficiently.

A Data Neural Network (DNN), on the other hand, has the advantages of high speed, high density, low power consumption, low cost, is self routing, and is a neural network. A DNN does not suffer the problems of traditional neural networks as communicating neurons need not be adjacent, and there is no distance limitation between them. Connectivity limitations do not exist in DNNs as the address space can be made as large as is needed. Unlike traditional data networks, DNNs are fundamentally secure. This is because Data Neurons establish a connection between each other using a non fixed address which is only known by the other neuron. DNNs build and rebuild connections automatically keeping connections up in a very reliable manner. Unlike the Internet, a DNN is a deterministic architecture with built in guarantees. Because a DNN is a neural network, it is intelligent. That intelligence can be used to solve complex math problems such as the Travelling Salesman Problem. A DNN can solve many logistic problems and can even be used to control the power grid.

Patent US 11,178,050 B2 teaches how to build a DNN, and also shows many application of this novel network. These applications are discussed in the patent and are identified by column and line:

  1. Secure packet data network, Col. 28, Line. 50
  2. Robotics, Col.29, Line. 44
  3. Sensor Networks, Col. 30, Line. 6
  4. Information Centric Networks, Col. 30. Line. 11
  5. Power Grid Control, Col. 30, Line. 40
  6. Mobile and Satellite Packet Data Networks, Col. 32, Line. 43
  7. Drawing Analysis: map, building, site development, schematic. Col. 27 Line. 18
  8. Logistics: optimized moving of assets in a complex network, Col. 28, Line. 4